druid forms at last!
July 29, 2010
I love the smell of data in the morning. It smells like… victory!
Druid forms; phew! It’s taken a while. The count of moonkin and tree forms was always straightforward since these forms derive from a talent. But now, thanks to all my correspondents on the question of feral druids and thanks to the power of modern datamining packages, I’m happy that I’ve got a reasonable estimate of those level 80 feral raiders who favour bear form and those who favour cat form.
The number of unclassifiable ferals is still a bit high but that is mainly due to the number of feral druids who do not raid. Only a tiny number of feral raiders really have an equal investment in cattish and bearish talents and glyphs. There may also be roles for cats and bears in PvP, but I don’t have a solution for estimating those.
Still we’ve got most of the answers we were after.
These numbers are for patch 3.3.3 and are the percentages based on all level 80 druids. So without further ado:
| Form | Popularity |
| Moonkin | 26% |
| Tree | 40% |
| Feral Raiders Cat-oriented | 11% |
| Feral Raiders Bear-oriented | 17% |
| Unclassifiable Ferals | 6% |
If we estimate cats and bears as a percentage of feral combat druids only, we get this:
| Form | Popularity |
| Feral Raiders Cat-oriented | 33% |
| Feral Raiders Bear-oriented | 50% |
| Unclassifiable ferals | 16% |
Percentages are based on active specs only, to keep things simple. That still strikes me as reasonable, over a large sample, since the percentages reflect what you’d see in-game on average.
I’ve got a couple more posts to come which explain in detail how the feral estimates are derived. And I’ll put up the data set so interested people can have a play around with it and see if there is any better way to cluster the bears and the cats. I’ll also add these tables to my druids reports over at the Google appengine site.
But that’s as good an estimate as I know how to get. And it was a fun ride getting to this point too.
July 29, 2010 at 8:43 pm
Great to see the mystery solved.
Sorry I, despite your motivation, did not write a post on this. But you have done so splendidly now!
If you want to relive the experience try to see if you can seperate DK tanks and DK DPS – as of now tanking is possible in all three trees, and so is DPS.
Personally I think the hardest part will be weeding out the “casuals” (Soloers) from the rest. But perhaps the lessons learned in this druid-frenzy can be applied.
Probably the easiet way would be to just look at the defence stat – but as always – my personal preference is to judge by talents and glyphs
July 30, 2010 at 10:50 am
Well, thanks to you for your insights – they were an enormous help in solving the problem. If you want to email me some name for yourself (eg your main character’s armoury page link, if you want to stay a bit anonymous) I’d like to add an acknowledgment of your contributions to the report that I post on the Google appengine site. DK’s would make an interesting datamining subject, for sure. We’ve certainly got the tools to study them now.
August 1, 2010 at 9:12 am
Nice job!
August 2, 2010 at 9:36 am
Hi Zardoz! I was very interested to see the culmination of your druid project.
I have a very simple query for you: can your coding skills tell us how many characters are Kingslayers? The WoW guild progression sites used to be a reliable means of estimating this, but now that enough players have it we are starting to get a lot of easier heroic mode kills done on characters that haven’t killed Arthas.
I’d be very interested to see, now that the 30% ICC buff is up, what percentage of level 80s have completed the 10-man and 25-man Lich King kills, how many of those have done it without getting the “downed all 12 bosses” achievement, and any other data you could dream up. Additionally, realm-by-realm data would be just awesome.
I’d be willing to help out with this, though I lack the technological expertise, if the project would require anything beyond a simple search by your software.
Cheers!
- GC
August 2, 2010 at 10:34 am
I’m not doing anything with achievements at the moment – I don’t collect the raw data, so there’s nothing to work with. They are interesting questions and I might get on to them, but not in the immediate future. Sorry.
October 15, 2010 at 11:42 pm
Hi, Nice post!
Doing my research I found great books about Data mining. This books intends to bring together the most recent advances and applications of data mining research in the promising areas of medicine and biology, in real life applications, web applications etc.
The readers will benefit from this books and consider it as an excellent way to keep pace with the vast and diverse advances of new research efforts.
This is link where you can find it: http://www.intechopen.com/search?q=Data+mining
They are free to download!!